Frontier AI Models Now Achieve Expert-Level Performance on Real Economic Work Tasks

The capability of frontier AI models to handle economically valuable professional work has entered a new phase, with recent benchmarks showing that cutting-edge systems now approach or match the deliverable quality of human experts across finance, legal research, software development, and taxation. New economic evaluation frameworks released in 2026 are providing clearer insights into AI's near-term impact on knowledge work, shifting focus from abstract capabilities to measured performance on tasks professionals actually perform.